The total variation distance between two double Wiener-Itô integrals (Q2435728)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 6259887
Language Label Description Also known as
default for all languages
No label defined
    English
    The total variation distance between two double Wiener-Itô integrals
    scientific article; zbMATH DE number 6259887

      Statements

      The total variation distance between two double Wiener-Itô integrals (English)
      0 references
      0 references
      19 February 2014
      0 references
      The author considers the total variation distance between two double Wiener-Itō integrals, improving upon a bound due to \textit{Yu. A. Davydov} and \textit{G. V. Martynova} [in: Statistics and control of stochastic processes. Moskva: Collect. Artic., 55--57 (1989; Zbl 0709.60056)] under an additional assumption concerning the eigenvalues of a certain Hilbert-Schmidt operator. Specifically, let \(H\) be a real, separable, infinite-dimensional Hilbert space. Let \(H^{\otimes2}\) and \(H^{\odot2}\) be the tensor product and symmetric tensor product, respectively, of \(H\) with itself. Let \(f_n\in H^{\odot2}\) be such that \(f_n\rightarrow f_\infty\in H^{\otimes2}\) as \(n\rightarrow\infty\). Let \(I_2(f_n)\) denote the double Wiener-Itō integral of \(f_n\) with respect to an isonormal Gaussian process on \(H\). Furthermore, let \(A_{f_\infty}:H\to H\) be given by \(A_{f_\infty}g=\langle f_\infty,g\rangle_H\) for \(f_\infty\in H^{\odot2}\) and let \(\lambda_{\infty,k}\), \(k\geq1\), denote the eigenvalues of \(A_{f_\infty}\). The author shows that, if \(|\{k:\lambda_{\infty,k}\not=0\}|\geq5\), then there is a constant \(c>0\) such that \[ \sup_C\left|\operatorname{P}(I_2(f_n)\in C)-\operatorname{P}(I_2(f_\infty\in C))\right|\leq c\| f_n-f_\infty\|_{H^{\otimes2}}\,. \] As an application, a rate is given for the convergence of a functional of two correlated fractional Brownian motions to the Rosenblatt distribution. The rate depends on the Hurst parameters of the fractional Brownian motions. The proofs use an approach developed by \textit{I. Nourdin} and \textit{G. Poly} [Stochastic Processes Appl. 123, No. 2, 651--674 (2013; Zbl 1259.60029)] which applies Malliavin calculus.
      0 references
      0 references
      double Wiener-Itō integrals
      0 references
      convergence in total variation
      0 references
      Malliavin calculus
      0 references
      isonormal Gaussian process
      0 references
      fractional Brownian motions
      0 references
      Rosenblatt process
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references